Biotin is an essential metabolite in all known free-living organisms.

Transcription

1 Host biotin is required for liver stage development in malaria parasites Teegan A. Dellibovi-Ragheb a, Hugo Jhun a, Christopher D. Goodman b, Maroya S. Walters a, Daniel R. T. Ragheb a, Krista A. Matthews a, Krithika Rajaram a, Satish Mishra a, Geoffrey I. McFadden b, Photini Sinnis a, and Sean T. Prigge a,1 a W. Harry Feinstone Department of Molecular Microbiology and Immunology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD 21205; and b School of BioSciences, University of Melbourne, Melbourne, VIC 3010, Australia Edited by Miguel Prudencio, Instituto de Medicina Molecular, Lisbon University, Portugal, and accepted by Editorial Board Member Stephen M. Beverley January 31, 2018 (received for review January 17, 2018) Acetyl-CoA carboxylase (ACC) is a biotin-dependent enzyme that is the target of several classes of herbicides. Malaria parasites contain a plant-like ACC, and this is the only protein predicted to be biotinylated in the parasite. We found that ACC is expressed in the apicoplast organelle in liver- and blood-stage malaria parasites; however, it is activated through biotinylation only in the liver stages. Consistent with this observation, deletion of the biotin ligase responsible for ACC biotinylation does not impede blood-stage growth, but results in late liver-stage developmental defects. Biotin depletion increases the severity of the developmental defects, demonstrating that parasite and host biotin metabolism are required for normal liver-stage progression. This finding may link the development of liver-stage malaria parasites to the nutritional status of the host, as neither the parasite nor the human host can synthesize biotin. acetyl-coa carboxylase apicoplast holocarboxylase synthetase biotin ligase Plasmodium Biotin is an essential metabolite in all known free-living organisms. It functions as a CO 2 carrier in carboxylation and decarboxylation reactions catalyzed by a small family of enzymes, the biotin carboxylases (1). Biotin carboxylases participate in central metabolic processes such as gluconeogenesis, lipogenesis, amino acid metabolism, and energy transduction (2). Bacteria, plants, and some fungi can synthesize biotin de novo from a pimeloyl-coa precursor, but other organisms acquire biotin from their diet or environment (3). Regardless of how biotin is acquired, a dedicated ATP-dependent biotin ligase, or holocarboxylase synthetase (HCS), is required to catalyze the covalent attachment of biotin to a specific lysine in the target carboxylase (4). Acetyl-CoA carboxylase (ACC) is the only protein predicted to be biotinylated in the parasite (5). ACC enzymes, the most common member of the biotin carboxylase family, are found in most living organisms and catalyze the first and rate-limiting step in fatty acid synthesis (FAS) (6). The Plasmodium ACC resembles the homomeric ACC found in the chloroplasts of grasses in that it contains three functional domains expressed as a single large polypeptide (7), and localizes to a plastid organelle called the apicoplast (8). The ACC orthologs that function in the plant chloroplast are common targets for herbicides (1). Some of these inhibitors have been tested against blood-stage malaria parasites (9), but it was shown that any growth inhibition observed with these compounds is the result of off-target effects, as Plasmodium falciparum ACC can be disrupted (8). ACC has two catalytic domains: a biotin carboxylase (BC) domain that transfers CO 2 from bicarbonate to biotin and a carboxyltransferase (CT) domain that transfers CO 2 from biotin to acetyl-coa to generate malonyl-coa. A third domain, known as biotin carboxyl carrier protein (BCCP), contains the conserved lysine to which biotin is attached. Biotin functions as a swinging arm to transfer the CO 2 moiety between the active sites of the BC and CT domains (1). The enzymatic domains of Plasmodium ACC have not yet been studied, but it has been shown that the BCCP domain of P. falciparum is biotinylated when expressed in Escherichia coli (10). Malaria parasites contain a type II FAS (FASII) pathway in the apicoplast (11) that relies on malonyl-coa as the two-carbon subunit for fatty acid elongation (12). Gene KOs of FASII pathway enzymes in the rodent parasites Plasmodium yoelii and Plasmodium berghei demonstrated that the FASII pathway is required for normal liver-stage development, but not for bloodor mosquito-stage development (13, 14). Thus, ACC and biotin should be required in the liver stages to provide malonyl-coa for downstream FASII pathway enzymes. Consistent with this idea, pyruvate dehydrogenase, which produces acetyl-coa, the substrate of ACC, has also been shown to be critical for liver-stage development in P. yoelii (15). Biotin metabolism may also be important for other stages of parasite development. Malaria parasites appear to encode enzymes comprising a fatty acid elongation (ELO) pathway similar to that described in the apicomplexan parasite Toxoplasma gondii (16). As is the case for typical ELO pathways (17), enzymes of the T. gondii ELO pathway are associated with the ER membrane and use malonyl-coa as a substrate (16). P. falciparum parasites lacking a functional FASII pathway can still elongate fatty acids, possibly because of the activity of the ELO pathway (14). If malonyl-coa can cross the apicoplast membranes, ELO as well as FASII could depend on ACC activity and biotin metabolism. Perhaps the most unusual feature of biotin metabolism in malaria parasites is the presence of two HCS paralogs encoded Significance Malaria parasites require certain host nutrients for growth and survival. In this project, we examined the role of the human vitamin biotin in all stages of the malaria life cycle. We cultured blood- and liver-stage malaria parasites in the absence of biotin and found that, whereas blood-stage replication was unaffected, liver-stage parasites deprived of biotin were no longer capable of establishing a blood-stage infection. Interestingly, biotin depletion resulted in more severe developmental defects than the genetic disruption of parasite biotin metabolism. This finding suggests that host biotin metabolism also contributes to parasite development. Because neither the parasite nor the human host can synthesize biotin, parasite infectivity may be affected by the nutritional status of the host. Author contributions: T.A.D.-R., H.J., C.D.G., M.S.W., K.A.M., and S.T.P. designed research; T.A.D.-R., H.J., C.D.G., M.S.W., D.R.T.R., K.A.M., K.R., S.M., and S.T.P. performed research; T.A.D.-R., H.J., C.D.G., M.S.W., D.R.T.R., K.A.M., K.R., G.I.M., P.S., and S.T.P. analyzed data; and T.A.D.-R., H.J., C.D.G., G.I.M., P.S., and S.T.P. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. M.P. is a guest editor invited by the Editorial Board. This open access article is distributed under Creative Commons Attribution-NonCommercial- NoDerivatives License 4.0 (CC BY-NC-ND). 1 To whom correspondence should be addressed. sprigge2@jhu.edu. This article contains supporting information online at /pnas /-/DCSupplemental. Published online February 26, E2604 E2613 PNAS vol. 115 no. 11

2 in the genomes of Plasmodium species. Plant cells also contain two HCS paralogs (18), and HCS activity is partitioned among the three compartments in which biotin-dependent carboxylases are found: the mitochondria, chloroplasts, and cytosol (19, 20). In Arabidopsis thaliana, alternative splicing governs the production of multiple AtHCS1 isoforms that are targeted to different compartments in the cell (21). Multiple isoforms of AtHCS2 are also produced, but these do not appear to have HCS activity, and deletion of the AtHCS2 gene has no observable phenotype in A. thaliana (21). We sought to determine the role of biotin across the malaria life cycle, including how biotin is acquired, the activity and localization of the biotin ligases, and the consequence of disrupting biotin metabolism in malaria parasites. Results ACC Is Located in the Apicoplast in Liver and Blood Stages but Is Not Biotinylated During the Blood Stages. ACC is the only predicted biotin-dependent enzyme in the Plasmodium genome. ACC has previously been localized to the apicoplast in blood-stage P. falciparum (8), but the localization has not been confirmed in the other stages of the parasite life cycle or in other Plasmodium species. We localized ACC in liver-stage P. berghei through immunofluorescence microscopy of infected HepG2 human liver cells by using antibodies directed against the BCCP domain of PfACC (10). We found that BCCP colocalizes with an apicoplast marker, the acyl carrier protein (ACP), indicating that endogenous PbACC resides in the apicoplast in liver-stage parasites (Fig. 1A). ACC enzymes are only active when covalently modified by biotin. To determine whether PbACC could be biotinylated in the liver stages, we performed immunofluorescence microscopy of HepG2 cells infected with P. berghei by using antibodies against biotin. We observed that biotin colocalizes with the apicoplast marker ACP, indicating that biotinylated proteins reside in the apicoplast in liver-stage malaria parasites (Fig. 1B). We next looked at ACC expression and biotinylation in bloodstage P. berghei by using immunofluorescence analysis with antibodies specific to BCCP and biotin. We found that PbACC is expressed in the apicoplast in blood-stage parasites (Fig. 1C), but we were unable to detect any biotinylated proteins in the blood stages (Fig. 1D). Thus, ACC is expressed in the apicoplast in blood- and liver-stage malaria parasites, but appears to be biotinylated only in the liver stages. To determine whether endogenous ACC is biotinylated in blood-stage P. falciparum, we performed immunofluorescence analysis of Dd2 attb parasites by using antibodies specific to BCCP and biotin. We could detect ACC in the apicoplast, and, in some experiments, ACC protein appeared to be proximal to the apicoplast as well (Fig. 1E). This phenomenon was also observed in P. berghei and may reflect slow kinetics of trafficking such a large protein ( 3,000 aa). We could not detect any biotinylated protein in P. falciparum because there was no significant signal above the background autofluorescence observed for uninfected red blood cells (Fig. 1F). These results indicate that, as in P. berghei, PfACC is expressed but not biotinylated in the blood stages, and raise the possibility that ACC activity is regulated through stage-specific biotinylation. To investigate the requirements for ACC trafficking to the apicoplast, we expressed the first 115 residues of PfACC (containing the putative targeting domain) fused to a C-terminal GFP tag in Dd2 attb parasites. We found that ACC-GFP localizes to an elongated, branched organelle distinct from the mitochondrion, which is consistent with apicoplast morphology (SI Appendix, Fig. S1A), and colocalizes with the apicoplast marker ACP (SI Appendix, Fig. S1B). We also observed that the fusion protein was processed normally after import into the apicoplast (SI Appendix, Fig. S1C). Taken together, these results demonstrate that the ACC leader peptide is sufficient to direct the protein to the apicoplast in blood-stage P. falciparum. Fig. 1. ACC is expressed in the apicoplast and is biotinylated only in liverstage parasites. (A) Immunofluorescence analysis of HepG2 cells at 48 hpi with P. berghei showing colocalization of PbACC, labeled with antibodies specific to the BCCP domain of ACC (αbccp), with the apicoplast marker αacp. The nuclei are labeled with DAPI. (B) Immunofluorescence analysis similar to A showing colocalization of biotinylated proteins (αbiotin) with the apicoplast marker αacp. (C) Immunofluorescence analysis of blood-stage P. berghei showing colocalization of PbACC (αbccp) with the apicoplast marker αacp. (D) Immunofluorescence analysis of blood-stage P. berghei showing no detectable biotinylated protein (αbiotin). (E) Immunofluorescence analysis of blood-stage P. falciparum showing colocalization of PfACC (αbccp) with the apicoplast marker αacp. (F) Immunofluorescence analysis of blood-stage P. falciparum showing no detectable biotinylated protein (αbiotin) compared with an uninfected control. (Scale bars: 5 μm.) HCS1 and HCS2 Localize Primarily to the Cytosol. We identified two biotin ligase paralogs (HCS1 and HCS2) in the Plasmodium genome. At least one of these enzymes should be responsible for catalyzing the biotinylation of PbACC; however, neither was predicted to localize to the apicoplast by targeting algorithms. To determine the subcellular location of the two biotin ligases in P. berghei, we tagged both genes with GFP through single cross-over recombination (SI Appendix, Fig. S2). Immunofluorescence analysis demonstrated that the GFP-tagged proteins are not predominantly located in the apicoplast in liver-stage or bloodstage parasites (SI Appendix, Fig. S3). Because of its size, the GFP tag may alter protein trafficking and subcellular location. To address this possibility, we generated HA-tagged parasite lines through double cross-over recombination (SI Appendix, Fig. S4). Consistent with the results from the GFP-tagged lines, the MICROBIOLOGY PNAS PLUS Dellibovi-Ragheb et al. PNAS vol. 115 no. 11 E2605

3 HA-tagged proteins do not have obvious apicoplast colocalization (Fig. 2). This was true at different time points during liverstage development (24, 48, and 56 h) and during blood-stage development (Fig. 2C). We also localized the two biotin ligase paralogs in blood-stage P. falciparum. We overexpressed a second copy of each ligase as a GFP fusion protein in Dd2 attb parasites and characterized these lines by using fluorescence microscopy of live cells labeled with DAPI. We found that both fusion proteins primarily localize to the cytosol (Fig. 3 A and B). We also performed Western blot analysis on whole-cell lysate from these parasites by using antibodies against GFP and detected a band of the correct mass for the full-length fusion protein in both biotin ligase overexpression lines (Fig. 3 C and D). Interestingly, PfHCS2-GFP was more difficult to detect than PfHCS1-GFP despite the fact that the expression of both proteins was driven by the same regulatory elements. This may be a result of a difference in protein stability between the two ligases. Together, these results demonstrate that the two biotin ligases are primarily cytosolic in the liver and Fig. 3. PfHCS1 and PfHCS2 are active biotin ligases. (A and B) Epifluorescence images of live P. falciparum blood-stage parasites expressing fulllength PfHCS1 fused to GFP (HCS1-GFP) or PfHCS2 fused to GFP (HCS2-GFP). The parasites were stained with Hoechst to identify nuclei. GFP fluorescence localizes to the cytosol in trophozoite (Top) and schizont (Bottom) parasites. (Scale bars: 5 μm.) (C and D) Western blot analysis of lysate from the transgenic parasites shown in A and B using antibodies against GFP (αgfp) showing a dominant band that corresponds to the expected size of the fulllength fusion protein. (E) Recombinant PfHCS1 and PfHCS2 were expressed in E. coli and purified by affinity chromatography, and their activity was tested by using an in vitro biotin ligase assay. PfHCS1, but not PfHCS2, biotinylated PfBCCP in vitro in an ATP-dependent reaction. (F) The activities of PfHCS1 and PfHCS2 were also tested in a cell-free assay. Lysates from parasites overexpressing the two ligases (shown in A and B), and a parental control, were incubated with purified BCCP under reaction conditions similar to E. PfHCS1 and PfHCS2 exhibit biotin ligase activity in this assay. blood stages of the parasite life cycle. As biotinylated proteins are located in the apicoplast of liver-stage parasites, this raises the question of whether the biotin ligase paralogs are active ligases and are responsible for protein biotinylation in the parasite. Fig. 2. HCS1 and HCS2 localize primarily to the cytosol in P. berghei. (A) Immunofluorescence analysis of HepG2 cells 24, 48, and 56 hpi with P. berghei parasites in which PbHCS1 was tagged with an HA tag through double cross-over recombination. The tagged proteins are labeled with specific antibodies (αha), the apicoplast is labeled with αacp, and the nuclei are labeled with DAPI. (B) Immunofluorescence analysis similar to that shown for A using P. berghei parasites in which PbHCS2 was tagged with an HA tag. (C) Immunofluorescence analysis of blood-stage parasites from the transgenic lines shown in A and B. (Scale bars: 5 μm.) PfHCS1 and PfHCS2 Are Active Biotin Ligases. To assess biotin ligase activity for PfHCS1 and PfHCS2, we first tested the activity of the two ligases in vitro. Recombinant PfHCS1 and PfHCS2 were expressed in E. coli and purified by affinity chromatography. The BCCP domain of PfACC was also expressed and purified as a biotinylation substrate. We found that overexpression of PfBCCP in E. coli results almost exclusively in apoprotein with less than 1% of the purified PfBCCP biotinylated. We tested the reactivity of each ligase toward PfBCCP in vitro and assessed PfBCCP biotinylation through affinity blotting with streptavidin-hrp followed by Western blotting with antibodies specific to PfBCCP. We found that PfHCS1, but not PfHCS2, biotinylated PfBCCP in an ATPdependent reaction (Fig. 3E). The very low signal in the absence E Dellibovi-Ragheb et al.

4 of ATP is probably the result of a small amount of biotinyl-amp copurified with PfHCS1. In addition, we performed streptavidin affinity-blot analysis of whole parasite lysate and observed a dramatic increase in total protein biotinylation in the PfHCS1 overexpression line (generated as described here earlier), which was not present in the PfHCS2 line or a parental control (SI Appendix, Fig. S5). This indicates that PfHCS1 has unusually lax substrate specificity, likely because of the release of the biotinyl-amp conjugate, which reacts spontaneously with proteins when it is free in solution. In a final approach, we performed cell-free assays with parasite lysate. Whole-cell extract was generated from parasites overexpressing PfHCS1 or PfHCS2, as well as the parental control, and the reactivity of these extracts toward recombinant PfBCCP was tested by using conditions similar to the in vitro biotinylation assay. In this assay, both ligases were active against PfBCCP (Fig. 3F), although PfHCS2 activity was slightly, but consistently, weaker than PfHCS1 in each repetition of this assay. The low level of biotinylation observed in the parental control is likely a result of endogenous biotin ligase activity. Taken together, these results show that malaria parasites express two functional biotin ligases, which represents an unusual example of an organism that contains two active biotin ligases. The fact that we observed activity for PfHCS2 only in the cell-free assay may be because PfHCS2 folds correctly only when expressed in P. falciparum or because this enzyme requires a binding partner not present in E. coli. Robust Biotinylation Activity Can Be Detected in the Cytosol of Blood- Stage Parasites. To determine whether blood-stage parasites are capable of biotinylating apicoplast proteins, we used the BCCP domain of PfACC as a genetic probe for biotinylation activity in P. falciparum. We expressed PfBCCP with a C-terminal GFP tag in Dd2 attb parasites. In the absence of any targeting sequence, the BCCP-GFP fusion protein would be expressed in the cytosol. We also expressed BCCP-GFP with an N-terminal apicoplasttargeting peptide, the first 55 residues of PfACP. We generated two constructs for each protein, one in which expression was driven by the strong calmodulin (CaM) promoter, and one in which expression was driven by the moderate ribosomal protein L2 (RL2) promoter (22). We were able to generate transgenic parasites from three of the constructs: the two cytosolic constructs and the apicoplast-localized protein with the RL2 promoter. The final construct failed perhaps because overexpression of apicoplast proteins by the CaM promoter is poorly tolerated a phenomenon we have observed before (23). We first verified that the GFP fusion proteins were correctly localized in each line by using live fluorescence microscopy (Fig. 4 A C) and immunofluorescence analysis (SI Appendix, Fig. S6). We then analyzed whole-cell lysate from the transgenic parasites by using affinity blotting with streptavidin-hrp and Western blotting with antibodies specific to GFP. We observed robust biotinylation of PfBCCP when it was targeted to the cytosol, but could not detect any biotinylation activity in the apicoplast in blood-stage parasites (Fig. 4 D F). The lack of biotinylation activity in the apicoplast agrees with the biotin ligase localization data, and eliminates the possibility of a cryptic or noncanonical biotinylation mechanism in the apicoplast. This also supports the conclusion that ACC is not biotinylated, and thus inactive, in parasite blood stages. Biotin Is Not Essential for the Survival of Blood-Stage Parasites. It has been previously demonstrated that the FASII pathway, which should be dependent on biotin, is important for the liver stages of the parasite lifecycle (13, 14). However, it is not known whether biotin is essential for parasite survival in the blood or mosquito stages. Geary et al. (24) treated blood-stage parasites with 0.2 mg/ml avidin and found that it had no effect on parasite growth in culture, but these experiments were inconclusive because they did not test whether avidin was able to permeate erythrocytes or measure Fig. 4. Robust biotinylation activity can be detected in the cytosol but not the apicoplast of blood-stage P. falciparum. (A) The biotinylation domain (BCCP) from PfACC was expressed as a GFP fusion protein in the apicoplast of blood-stage P. falciparum by using the apicoplast targeting domain from PfACP (ACP-BCCP-GFP) under control of the ribosomal L2 (RL2) promoter. Localization was verified by epifluorescence microscopy in late ring (Top), trophozoite (Middle), and schizont (Bottom) parasites. Parasites were stained with MitoTracker to identify the mitochondrion and DAPI to identify nuclei. (B) BCCP was expressed as a GFP fusion protein in the cytosol (BCCP- GFP) under control of the ribosomal L2 (RL2) promoter, and localization was verified by epifluorescence microscopy as in A. (C) BCCP was expressed as a GFP fusion protein in the cytosol (BCCP-GFP) under control of the stronger CaM promoter, and localization was verified by epifluorescence microscopy as in A. (D F) Whole-cell lysates from the parasite lines shown in A C were analyzed by Western blotting with antibodies against GFP and affinity blotting with streptavidin-hrp. BCCP is biotinylated in the cytosol but is not biotinylated when expressed in the apicoplast. (Scale bars: 5 μm.) the concentration of intraerythrocytic biotin stores. To determine whether blood-stage P. falciparum is auxotrophic for biotin, we measured parasite growth in biotin-depleted medium. The medium was prepared from biotin-free RPMI supplemented with the serum substitute AlbuMAX II. We tested the growth of Dd2 attb parasites as well as the three lines overexpressing the BCCP domain of PfACC (generated as described here earlier). To confirm that we had completely removed biotin from the cultures, we used PfBCCP as a sentinel protein to detect intracellular biotin. It should also function as a biotin sink to further sensitize parasites to biotin deprivation because it would compete with endogenous substrates for any trace amounts of biotin that remained. We found that the presence or absence of biotin in the media made no significant difference in parasite growth (Fig. 5A). On the eighth MICROBIOLOGY PNAS PLUS Dellibovi-Ragheb et al. PNAS vol. 115 no. 11 E2607

5 Fig. 5. Biotin is not essential for survival of blood-stage P. falciparum. (A) Blood-stage P. falciparum cultures were maintained for 8 d in media with or without biotin. Four parasite lines were used: the parental control (Dd2), parasites overexpressing BCCP in the apicoplast under the ribosomal L2 (RL2) promoter, and parasites overexpressing BCCP in the cytosol under the RL2 promoter or the stronger CaM promoter. Error bars represent SEM of three biological replicates. (B) On day 8 of the growth assay shown in A, parasites were harvested and biotinylation levels were assessed by streptavidin-hrp affinity blotting, followed by Western blotting with αbccp antibodies. There was no detectable biotinylated BCCP under biotin-free conditions, indicating that biotin was completely depleted in these cultures. Controls were parasites cultured in complete medium. (C) Blood-stage parasites (Dd2) were cultured with or without biotin for 7 d in media containing lipid-free BSA reconstituted with 30 μm palmitic acid and 30 μm oleic acid. Error bars represent SEM of three biological replicates. (D) TLC of parasites labeled with 14 C-acetate showing incorporation into C16 and C18 fatty acids. This activity was seen in WT and FASII KO parasites, and was independent of biotin. Similar results were obtained when this experiment was repeated (Bottom) using the minimal fatty acid medium described in C. No activity was seen with the negative control, uninfected red blood cells (RBC). day of the growth assay, parasites were isolated, and whole-cell extract was analyzed by affinity blotting with streptavidin-hrp followed by Western blotting with antibodies specific to PfBCCP. There was no detectable biotinylated PfBCCP in the cultures grown under biotin-free conditions, indicating that biotin was completely depleted from the medium (Fig. 5B). We therefore concluded that biotin is not essential for blood-stage replication in P. falciparum. In addition, these data show that parasites scavenge biotin rather than synthesize it, as cytosolic PfBCCP is biotinylated only when biotin is added to the medium. It has previously been shown that blood-stage malaria parasites have the ability to elongate fatty acids taken up from the medium (25). In addition, it has been demonstrated that parasites lacking a functional FASII pathway can still incorporate radiolabeled acetate into fatty acid species, most likely because of the activity of the ELO pathway (14). Based on these results, we tested whether parasites can survive without biotin under limited fatty acid culture conditions. We cultured parasites in serum-free medium supplemented with lipid-free BSA reconstituted with only palmitic acid (C 16:0 ) and oleic acid (C 18:1 ), which are the minimal requirements for parasite growth (25). Parasites cultured with or without biotin were maintained in parallel, and biotin made no difference in the growth rates of the parasites (Fig. 5C). These results demonstrate that biotin is not required for blood-stage parasite growth, even under conditions in which the FASII and ELO pathways would be needed the most. Biotinylated Proteins Cannot Be Detected in Blood-Stage Parasites. Thus far, we have been unable to show any biotin-dependent phenotype in blood-stage parasites. This raises the question of whether there are any biotinylated proteins in the blood stages. We were unable to detect any signal above background in immunofluorescence analysis of blood-stage P. falciparum (Fig. 1F) or P. berghei (Fig. 1D) by using antibodies specific to biotin. Similarly, we were unable to detect biotinylated proteins in the blood stages by using streptavidin-fitc (SI Appendix, Fig. S7A). Because these data do not rule out the possibility that there are biotinylated proteins present at a level below the limit of detection by microscopy, we used more sensitive techniques to probe this question further. In radiolabel uptake experiments with bloodstage P. falciparum, we did not detect incorporation of [ 3 H]biotin or [ 35 S]biotin into cellular proteins (SI Appendix, Fig. S7B), even though we have shown that the parasites are capable of scavenging and incorporating exogenous biotin (Fig. 5B). In a final approach, we performed affinity-blot analysis of parasite lysate by using streptavidin-hrp. Although we detected biotinylated proteins in some of the blots, these bands can often be detected in samples of uninfected red blood cells (SI Appendix, Fig. S7C). These three proteins correspond in mass to the human biotin carboxylases methyl-crotonyl-coa carboxylase (73.7 kda), pyruvate carboxylase (127.2 kda), and the ACC isoforms ( kda). Although there are no biotin carboxylases expressed in mature red blood cells, they are expressed in reticulocytes and leukocytes, which can be present in small amounts as contaminants in our culture system. Thus, if blood-stage malaria parasites contain any biotinylated proteins, these proteins are much less abundant than the trace levels of human proteins in our culture system. Biotin- and FASII-Independent Fatty Acid Elongation Activity Is Present in Blood-Stage P. falciparum. The growth of blood-stage P. falciparum parasites does not require biotin or ACC (8), leading us to the hypothesis that the ELO pathway may not rely on malonyl-coa produced by ACC. We therefore tested whether ELO activity could be detected under biotin-free conditions. Synchronized ring-stage parasites were incubated with [ 14 C]-acetate with or without biotin, after which free fatty acids were extracted and analyzed by thinlayer chromatography (TLC). To ensure we were assessing ELO activity, we compared WT (i.e., Dd2) parasites with FASII-KO (ΔFabI) parasites (14). We detected radiolabeled C 16 and C 18 fatty acids in both parasite lines cultured with or without biotin (Fig. 5D). Similar results were obtained in minimal fatty acid culture medium (Fig. 5D), conditions previously shown to elevate FASII activity (26). Thus, blood-stage parasites are capable of elongating fatty acids independent of FASII and biotin. Neither PbHCS1 Nor PbHCS2 Is Required for Blood-Stage Replication or Mosquito-Stage Development. To determine which biotin ligase is responsible for ACC biotinylation, both enzymes were knocked out in P. berghei ANKA parasites. The KO constructs were designed by using the pdef-hdhfr plasmid, which facilitates double cross-over homologous recombination, resulting in excision of the entire ORF of the target gene (SI Appendix, Fig. S8 A and C). Correct integration of the hdhfr cassette and loss of HCS1 and HCS2 were verified in independent clones by diagnostic PCR (SI Appendix, Fig.S8 B and D). The viability of both KO strains demonstrated that neither gene is essential for survival in the blood stages. To characterize development through the mosquito stages, WT and ΔHCS1 or ΔHCS2 gametocytes were fed to Anopheles stephensi mosquitoes, and infection levels were measured at different time points. Mosquito midgut oocysts were visualized 10 d after feeding, midgut sporozoites were counted 12 and 14 d after feeding, and salivary-gland sporozoites were counted between 16 and 20 d after feeding. Throughout mosquito stage development, ΔHCS1 and ΔHCS2 parasites exhibited no defects in oocyst formation, sporozoite development, or sporozoite E Dellibovi-Ragheb et al.

6 invasion of the salivary glands compared with WT controls (SI Appendix, Fig.S9). Loss of PbHCS1 Affects Early Liver-Stage Development. We next evaluated in vitro liver-stage growth and morphology of ΔHCS1 and ΔHCS2 parasites by using immunofluorescence analysis of infected HepG2 cells at different time points during liver-stage development. Early liver-stage morphology was visualized by using mab 3D11, which recognizes the circumsporozoite protein (CSP) on the parasite plasma membrane (27), at 24 and 48 h post infection (hpi; SI Appendix, Fig. S10 A and B). We also used antibodies specific for a parasitophorous vacuole protein, UIS4 (28), at 48 hpi (SI Appendix, Fig. S10C). At 24 hpi, ΔHCS1 parasites were an average of 3 μm 2 smaller than ΔHCS2 parasitesorwtcontrols(si Appendix, Fig. S10D). At 48 hpi, ΔHCS1 parasites were an average of 53 μm 2 smaller than WT parasites (SI Appendix, Fig. S10E). Taken together, these data indicate that parasites lacking either of the biotin ligases can progress through early liver-stage development, but that ΔHCS1 parasites display a growth phenotype. Deletion of PbHCS1, but Not PbHCS2, Results in the Loss of Liver- Stage Protein Biotinylation. To assess liver stage biotinylation in the biotin ligase deletion lines, immunofluorescence analysis was performed on HepG2 cells at 48 hpi with WT, ΔHCS1, orδhcs2 parasites. PbACC expression was visualized by using antibodies against the BCCP domain of PfACC, verifying that the enzyme is expressed normally in both deletion lines (Fig. 6A). Although we did not detect any change in biotinylation in ΔHCS2 parasites, we observed a complete loss of protein biotinylation in ΔHCS1 parasites (Fig. 6 B and C). These data indicate that PbHCS1 is essential for protein biotinylation in liver-stage parasites. Parasites Lacking PbHCS1 Show Delayed Progression Through Liver Stages in Vivo. We measured the prepatent period of the biotin ligase deletion lines by inoculating mice with salivary-gland sporozoites and measuring the length of time before bloodstage parasites were detected microscopically. We found that i.v. injection of 500 sporozoites per mouse did not always result in blood-stage infection, and that this was particularly true for the ΔHCS1sporozoites (SI Appendix, Table S1). Increasing the inoculum to 5,000 sporozoites per mouse reproducibly produced a blood-stage infection and revealed an average delay in patency of 2.1 d for ΔHCS1 parasites compared with WT controls. By contrast, ΔHCS2 parasites were indistinguishable from WT in the number of mice that achieved a patent blood-stage infection and the prepatent period. Thus, deletion of PbHCS1, but not PbHCS2, significantly impairs the ability of malaria parasites to progress from the liver stages to the blood stages. Deletion of PbHCS1 Affects Late Liver-Stage Development and Reduces Merosome Production. We visualized merozoite morphology by immunofluorescence microscopy at the late liverstage time point of 56 h. By using antibodies specific for MSP1 (merozoite surface protein 1), we generally observed reduced numbers of merozoites in ΔHCS1 parasites compared with WT or ΔHCS2 cells (Fig. 7A). Consistent with this observation, the size of the parasitophorous vacuole (assessed by using antibodies specific for the parasitophorous vacuole protein UIS4) appeared to be smaller in ΔHCS1 parasites (Fig. 7B). On average, ΔHCS1 parasites were approximately 73 μm 2 smaller at the 56-h time point, whereas no significant size decrease was observed in ΔHCS2 cells (Fig. 7C). We next measured the number of merosomes (i.e., membrane-bound clusters of merozoites) released from HepG2 cells infected with WT or ΔHCS1 parasites at 65 h post invasion. From independent experiments, each performed in triplicate, we found that fewer merosomes were produced by ΔHCS1 parasites compared with WT controls (Table 1). Taken together, these results show that deletion of PbHCS1 results in Fig. 6. Deletion of PbHCS1, but not PbHCS2, results in the loss of liver-stage protein biotinylation. (A) Immunofluorescence analysis of HepG2 cells at 48 hpi with WT, ΔHCS1,orΔHCS2 P. berghei showing that PbACC (αbccp) is expressed in the apicoplast (αacp) in both deletion lines. (B) Immunofluorescence analysis similar to A showing that biotinylated protein (αbiotin) can be detected in the apicoplast (αacp) in WT and ΔHCS2 parasites, but not ΔHCS1 parasites. Regions of protein biotinylation surrounding the parasite are presumably host mitochondria, which are thought to be recruited to the parasitophorous vacuole (52). (Scale bars: 5 μm.) (C) Biotin signal associated with the apicoplast and nonapicoplast compartments is expressed as a percentage of the total biotin signal found in the host cell. Error bars represent SEM from four or more independent experiments (ΔHCS1, n = 30, n = 6 replicates; ΔHCS2, n = 13, n = 4 replicates; WT, n = 22, n = 6 replicates). ns, not significant; ***P < late liver-stage defects, and suggests that biotin metabolism is important for this stage of development. Host Biotin Is Required for Normal Liver-Stage Development. We then tested whether parasites require biotin scavenged from the host to complete liver-stage development. We cultured HepG2 cells by using media that had been dialyzed against avidin to remove biotin (which is present in serum) and then supplemented with additional avidin. Importantly, this biotin-depletion medium did not affect hepatocyte morphology or survival over a 6-d test period (SI Appendix, Fig. S11). Cells were seeded and maintained in biotin-depleted media or normal media and infected with MICROBIOLOGY PNAS PLUS Dellibovi-Ragheb et al. PNAS vol. 115 no. 11 E2609

7 less in the ΔHCS1 parasites, which, when combined with the lower number of merosomes produced by this mutant, results in a 40-fold overall decrease in the number of merozoites produced. This could account for the significant difference in the prepatency assays (SI Appendix, Table S1). Under biotin-depleted conditions, fewer merozoites were produced by WT and ΔHCS1 parasites than were produced by the mutant when grown in the presence of biotin (Table 1). We assessed the effect of biotin depletion on infected HepG2 cells with immunofluorescence analysis and confirmed that host and parasite biotinylation was significantly reduced (SI Appendix, Fig. S12). Together, these data indicate that liver-stage parasites rely on biotin scavenged from the host, and that biotin is required for normal liver-stage development. These data also suggest that perturbation of biotin metabolism in the host cell may further impede the development of liver-stage parasites. Biotin Is Required for Merosome Infectivity. We assessed the ability of merosomes produced in vitro to initiate blood-stage infections in mice. Merosomes were produced as described earlier in HepG2 cells cultured in biotin-depleted media or normal media. Across three independent experiments, merosomes produced in normal medium were usually infectious (14 of 14 mice for WT and 9 of 13 mice for ΔHCS1) with a delay to patency for ΔHCS1 similar to that observed in SI Appendix,TableS1. However, WT merosomes produced under biotin-deficient conditions were infectious in only 1 of 14 mice. Quantification of total merozoites indicates that the reduced infectivity was not caused by lower numbers of merozoites produced (Table 2). Mutant parasites cultured in biotindeficient medium were never infectious. When combined with the data presented in Table 1, these experiments show that biotin deficiency affects the infectivity of merozoites as well as reducing the total number of merosomes produced. Discussion We identified one biotinylation substrate (ACC) in malaria parasites, but two biotin ligases (HCS1 and HCS2). This unusual arrangement seems to be conserved in all species of malaria parasites; however, the organization of biotin metabolism is quite divergent among different apicomplexan genera. The apicomplexan Cryptosporidium parvum does not contain an apicoplast and encodes a cytosolic ACC and cytosolic biotin ligase. By contrast, Theileria annulata lacks both ACC and biotin ligases. In T. gondii, two biotinylated proteins (ACC and pyruvate carboxylase) were Fig. 7. Liver-stage development is abnormal in ΔHCS1 parasites. (A) Immunofluorescence analysis of infected HepG2 cells 56 hpi with WT, ΔHCS1, or ΔHCS2 P. berghei. The parasite plasma membrane is illuminated with the late liver-stage marker αmsp1, and the nuclei are labeled with DAPI. (B) Immunofluorescence analysis similar to A at 56 hpi by using antibodies specific for the parasitophorous vacuole protein UIS4. (C) Quantification of cell area at 56 hpi from three biological replicates with at least 20 measurements per parasite strain. (Scale bars: 5 μm.) ns, not significant; ****P < WT or ΔHCS1sporozoites, and merosomes were collected and counted 65 h post invasion. At least five replicates were used for each condition. Under normal biotin-replete conditions, we found that there was an approximately fourfold reduction in the number of merosomes produced by ΔHCS1 parasites compared with WT (Table 1); however, the size of the ΔHCS1 merosomes appeared consistently smaller than those of WT parasites. We therefore quantified total merozoite numbers by using quantitative PCR and found that ΔHCS1merosomes contain fewer merozoites. We observed that the number of merozoites per merosome was 10-fold Table 1. Host biotin is required for normal liver stage development Expt Line Biotin Mrsm per well (SD) Mzt/Mrsm Mzt per well WT + 6,900 (1,924) ΔHCS1 + 1,550 (788) 2 WT + 17,390 (5,663) 3,307 58,000 WT 670 (303) ΔHCS1 + 3,270 (434) ΔHCS1 2,330 (1,492) WT + 1,500 (418) 4,867 7,300 WT 313 (314) ΔHCS (267) ΔHCS1 250 (194) WT + 3,000 (576) 3,667 11,000 WT 438 (304) ΔHCS (359) ΔHCS1 250 (188) Expt, experiment number; Mrsm, merosome; Mzt, merozoite. E Dellibovi-Ragheb et al.

8 Table 2. Biotin is required for merosome infectivity Expt Line Biotin Mrsm injected Mice positive P0, d Mzt injected WT + 1,000 4/4 3.5 WT 1,000 0/4 * ΔHCS1 + 1,000 3/3 5.7 ΔHCS1 1,000 0/1 * 2 WT + 2,500 5/ ,000 WT 2,500 0/5 * 520 ΔHCS1 + 2,500 2/ ΔHCS1 2,500 0/5 * WT + 2,000 5/ ,700 WT 2,000 1/ ΔHCS1 + 2,000 4/ ΔHCS1 2,000 0/5 * 160 Expt, experiment number; Mice positive, fraction of mice infected with blood stage parasites; Mrsm, merosome; Mzt, merozoite; P0, prepatent period. *Parasites were not detected 1 mo after merosome injection. identified (29), but there is only one predicted biotin ligase (TGME49_089760). The duplication of biotin ligases in malaria parasites is similar to the situation found in the plant A. thaliana. The two HCS genes in A. thaliana share 71% sequence identity and likely resulted from a gene duplication event. Similarly, HCS1 and HCS2 from P. falciparum share 47% sequence identity and are likely paralogs created by gene duplication. AtHCS1 is essential for viability, whereas disruption of the AtHCS2 gene does not lead to any observable phenotype (21). Similarly, deletion of P. falciparum HCS1 results in a growth defect whereas deletion of HCS2 does not (SI Appendix, TableS1). It is possible that HCS2 and AtHCS2 share a common role that has yet to be determined. Both biotin ligase paralogs are predominantly located in the cytosol in the blood stages of P. berghei (Fig. 2C) and P. falciparum (Fig. 3 A and B). It was anticipated that at least one of the ligases would be located in the apicoplast because this subcellular compartment was thought to contain ACC (30, 31). Indeed, ACC is located in the apicoplast in P. berghei and P. falciparum (Fig. 1 C and E), consistent with apicoplast localization of ACC in the related apicomplexan parasite T. gondii (29). Although ACC is biotinylated in T. gondii (29), it is not biotinylated in blood-stage malaria parasites (Fig. 1 D and F). We found this result to be surprising and used a genetic probe to determine whether any biotin ligase activity is located in the apicoplast. When we expressed the BCCP domain of ACC in blood-stage P. falciparum parasites, we found that the probe was biotinylated when expressed in the cytosol, but that there was no detectable biotinylation when it was expressed in the apicoplast (Fig. 4). This result is consistent with the cytosolic localization of both biotin ligases and indicates that the apicoplast of blood-stage parasites does not contain any enzymes with biotin ligase activity. Taken together, these results show that blood-stage malaria parasites express ACC and two active biotin ligases, but ACC is not biotinylated because all biotin ligase activity is confined to the cytosol whereas ACC is exclusively located in the apicoplast. The fact that ACC is not biotinylated in blood-stage malaria parasites suggests that biotin is not needed during this stage of parasite development. We cultured P. falciparum parasites expressing the BCCP domain in biotin-depleted growth medium. These parasite lines grew equally well regardless of whether biotin was added back to the medium (Fig. 5A). Subsequent analysis of parasites expressing BCCP in the cytosol demonstrated that BCCP was biotinylated only when biotin was added back to the depleted medium (Fig. 5B). These experiments show that blood-stage parasites do not synthesize biotin and do not rely on this vitamin for growth, but can import external biotin when it is supplied in the medium. Even when parasites were cultured in growth medium designed to increase the reliance on fatty acid metabolism, there was still no growth phenotype associated with biotin depletion (Fig. 5C). The ability of blood-stage malaria parasites to grow without biotin has interesting consequences for fatty acid metabolism. All known FAS and ELO pathways use malonyl-coa for chain elongation. ACC is the only predicted source of malonyl-coa in malaria parasites, and thus should be required for FASII and ELO activity. Key enzymes of the FASII pathway, such as FabI, have been knocked out in human and rodent malaria species, demonstrating that FASII is dispensable for blood-stage replication (13, 14). The lack of ACC biotinylation in blood-stage parasites further suggests that FASII cannot function during this stage of parasite development. Fatty acid elongation, however, can still be observed in ΔFabI parasites, implying that the ELO pathway is active in the blood stages of P. falciparum and P. berghei (14). We used the P. falciparum ΔFabI line to show that radiolabeled acetate is still incorporated into parasite fatty acids regardless of whether biotin is present in the growth medium, and, furthermore, the incorporation of acetate was the same between ΔFabI parasites and the parental line (Fig. 5D). Taken together, these data suggest that FASII is inactive in blood-stage P. falciparum parasites whereas these parasites have ELO activity that does not depend on biotin or ACC. The ELO pathway could rely on another source of malonyl-coa, such as a malonyl-coa synthetase (32). In liver-stage malaria parasites, the arrangement of ACC and the biotin ligases appears to be the same as it is in blood stages. The ACC is located in the apicoplast (Fig. 1) whereas the biotin ligases are predominantly located in the cytosol (Fig. 2 and SI Appendix, Fig. S3). The obvious difference is that apicoplast proteins (presumably ACC, although there may be others) are biotinylated, and this raised the question of which ligase is responsible for this activity. To answer this question, we deleted the genes encoding HCS1 and HCS2 in P. berghei and found that HCS1 is exclusively responsible for protein biotinylation whereas deletion of HCS2 has no detectable liver-stage phenotype (Fig. 6). This result suggests that HCS1 is partially located in the apicoplast of liver-stage parasites, albeit at low levels compared with the majority of HCS1 located in the cytosol (Fig. 2A). There are examples in other organisms of dual targeting of biotin ligases to different cellular compartments, which is regulated by alternative splicing of a single mrna or alternative translation initiation sites. This has been shown in some detail in A. thaliana (21) and has been suggested in Pisum sativum (20) as well as humans (33). In addition, there are examples of dual targeting of proteins to the apicoplast, mitochondrion, and cytosol in malaria parasites (34 36), although the mechanisms are not well understood. One suggested mechanism is the use of an ambiguous targeting sequence that is inefficiently recognized by the protein sorting machinery, so that a fraction of the total protein is imported into an organelle and the remainder is retained in the cytosol (37). Regardless of the mechanisms involved, our results show that biotinylation by HCS1 is stage-specific, occurring in liver-stage but not blood-stage malaria parasites. P. berghei parasites lacking HCS1 have a defect in liver-stage development that begins early in parasite development (SI Appendix, Fig. S10) and becomes more obvious by 56 h post invasion (Fig. 7). ΔHCS1 parasites produce fewer merosomes than WT parasites, and these merosomes contain significantly fewer merozoites (Table 1). In general, late liver-stage developmental defects and severely curtailed formation of merosomes are hallmarks of KO parasites with FASII gene deletions (13, 14) or deletions of genes required for the FASII pathway (15, 38). Deletion of FASII-related genes in P. berghei [ΔFabI (14) and ΔLipB (38)] results in a 4-d delay to patency, with few parasites able to complete liver-stage development and almost no merosome formation. By contrast, ΔHCS1 parasites display a 2.1-d delay to patency and MICROBIOLOGY PNAS PLUS Dellibovi-Ragheb et al. PNAS vol. 115 no. 11 E2611

9 quantifiable merosomes. If HCS1 is responsible for activating ACC to supply malonyl-coa for FASII, why is the ΔHCS1 phenotype different from that of ΔFabI or ΔLipB? One explanation for the less severe ΔHCS1 phenotype is that the parasites have another source of malonyl-coa, allowing them to partially bypass the need for HCS1. Hepatocyte ACC produces malonyl-coa, and it is possible that this metabolite can be scavenged. This explanation is consistent with the observation that WT and ΔHCS1 parasites produce fewer merozoites when cultured in biotindepleted medium (Table 1). Presumably, the deletion of HCS1 affects only parasite ACC activity, whereas biotin depletion affects parasite and host cell ACC activity. To assess the combined effects of host cell and parasite biotin deprivation, we conducted merosome infection experiments. We found that WT and mutant ΔHCS1merosomes produced in vitro by host cells cultured in complete medium were usually infectious in mice, but were almost never infectious if they were produced in biotin-deficient conditions (Table 2). The dramatic decrease in infectivity could result from the merozoites failing to egress from the membrane-bound merosomes, or it could be that the merozoites are defective. It is difficult to distinguish between these possibilities in our experiments; however, the fact that many merozoites are released during the injection process suggests that the merozoites themselves are less infectious. These results raise the possibility that biotin is a host nutrient that could affect parasite development, as has been suggested for other nutrients (39). In humans, biotin deficiency is almost invariably caused by mutations in an essential enzyme called biotinidase, which is required to make use of biotin obtained from dietary sources (40). More than 165 biotinidase mutations have been described in humans (41). In some cases, these mutations are severe enough to cause disease, but it is more common for these mutations to decrease biotin utilization without causing disease. Although the data currently available are not sufficient to implicate biotin deficiency in protection from malaria infection, there is some evidence that biotinidase mutations are more prevalent in certain African populations (42). In conclusion, we found that ACC is expressed in the apicoplast organelle in liver- and blood-stage malaria parasites, but it is biotinylated only in the liver stages, suggesting that ACC activity is regulated through stage-specific biotinylation. Because ACC is the first committed step in FASII pathways, protein biotinylation may serve as a molecular switch controlling fatty acid biosynthesis in malaria parasites. Gene deletion experiments showed that the biotin ligase HCS1 is solely responsible for biotinylation in liverstage parasites. Parasites lacking HCS1 show a late liver-stage defect and produce fewer merosomes with abnormally low numbers of merozoites per merosome. Biotin depletion results in more severe developmental defects, resulting in merozoites that are no longer capable of blood-stage infection. These results imply that host and parasite biotin metabolism are required for normal liverstage progression and merozoite infectivity. Thus, the development of liver-stage malaria parasites may be linked to the nutritional status of the host, as neither the parasite nor the human host can synthesize biotin. Materials and Methods Cells and Antibodies. HepG2 human liver carcinoma cells [HB-8065; AmericanType Culture Collection (ATCC)] were maintained in Dulbecco s modified Eagle medium (DMEM) supplemented with 10% FCS, 2 mm glutamine, and 1 penicillin streptomycin. The following antibodies were used: mouse monoclonal α-biotin (no ; Jackson) at 1:250 or 1:500 dilution; rabbit BCCP antiserum (10) (raised against the P. falciparum antigen but recognizes P. falciparum and P. berghei ACC) at 1:100 (Pf) or 1:250 (Pb) dilution; rabbit or rat polyclonal αacp (43, 44) (raised against the P. falciparum antigen but recognizes P. falciparum and P. berghei ACP) at 1:500 (Pf) or 1:1,000 (Pb) dilution; mouse monoclonal αcsp 3D11 (27) at 1:1,000 dilution; rabbit αuis4 (28) at 1:1,000 dilution; P. yoelii MSP1-19 rabbit antiserum (MRA-23, MR4; ATCC) at 1:1,000 dilution; and mouse monoclonal αgfp (Roche) at 1:200 (Pf) or 1:1,000 (Pb) dilution. The corresponding secondary antibodies conjugated to Alexa Fluor-488 or Alexa Fluor-594 (Invitrogen) were used as indicated. Immunofluorescence Analysis. To visualize liver-stage parasites, HepG2 cells were seeded at a density of 10 5 cells per well on Permanox eight-well chamber slides coated with collagen (no ; BD Biosciences). P. berghei ANKA sporozoites were harvested from infected Anopheles stephensi mosquitoes d post blood meal, quantified, and centrifuged onto the HepG2 cells at a density of sporozoites per well. At the time point indicated, cells were fixed with 4% paraformaldehyde for 20 min at room temperature, washed with PBS solution, and permeabilized with methanol overnight at 20 C. Slides were blocked and probed in 1% BSA in PBS solution and mounted with Prolong Gold Antifade Reagent with DAPI (Invitrogen). Average cell area was quantified by using Volocity software (PerkinElmer). Immunofluorescence analysis of P. berghei and P. falciparum blood-stage parasites was carried out as previously described (45) with a few modifications (46). The SI Appendix provides more details. P. falciparum Transfections. Parasite transfections were carried out by using the mycobacteriophage Bxb1 integrase system developed by Nkrumah et al. (47) and modified by Spalding et al. (48). Vector construction using the primers listed in SI Appendix, Table S2 is described in the SI Appendix. Endogenous Tagging and Localization of P. berghei Biotin Ligases. Double cross-over replacement constructs were made for the P. berghei biotin ligase genes HCS1 (PBANKA_051100) and HCS2 (PBANKA_132360). P. berghei ANKA genomic DNA was amplified and cloned into a vector designed to append a 3xHA tag to the 3 end of the ligase genes. A single cross-over strategy was used to generate GFP-tagged parasite lines (SI Appendix). Constructs were linearized and transfected into P. berghei ANKA according to Janse et al.(49). Biotin Ligase Activity Assays and Affinity/Western Blot Analysis. The biotin ligase genes were subcloned into expression vector pmalcht (50) for expression and purification as MBP fusion proteins. The truncated BCCP domain of PfACC was expressed as a GST fusion protein in BL21-Star(DE3) cells by using a modified pgex-4t-3 expression vector (pgext) and purified as previously described (10). The biotin ligase in vitro assay was carried out by using pure recombinant BCCP and biotin ligase proteins in the presence of ATP and biotin. Biotinylated BCCP was detected by affinity blot using streptavidin coupled to peroxidase (ultrasensitive; Sigma), and the presence of BCCP protein was confirmed by anti-bccp Western blot. For the cell-free assay, P. falciparum parasites were isolated by saponin lysis followed by sonication to release parasite proteins. BCCP, biotin, and ATP were added to the parasite lysates, and biotinylation activity was assessed by affinity and Western blot. Growth Assay and Radiolabel Incorporation. Biotin-depleted medium was prepared using RPMI medium 1640 modified with L-glutamine without phenol red and biotin (R ; US Biological), which was supplemented with 25 mm Hepes, 0.375% sodium bicarbonate, 12.5 μg/ml hypoxanthine, 5 g/l AlbuMAX II, and 25 μg/ml gentamicin. For each parasite line, the cultures and red blood cells were washed twice in 10 ml biotin-free RPMI to remove biotin and then divided into parallel cultures for growth assays conducted with or without biotin supplementation. Lipid-free BSA (Sigma) was reconstituted with palmitic acid (Sigma) and potassium oleate (Fluka) and used in lieu of AlbuMAX for growth assays under minimal lipid conditions. For radioactive biotin incorporation experiments, P. falciparum cultures were maintained with daily feedings of biotin-free medium. Parasite cultures were labeled for 3 d with 0.48 nmol 3 H-biotin (American Radiolabeled Chemicals) or 29.2 nmol of freshly prepared 35 S-biotin (10) each day for 3 d. Label incorporation was assessed by autoradiography. For acetate incorporation, synchronized ringstage parasites were incubated for 24 h with 10 μci/ml 14 C-acetate, after which parasites were isolated by saponin lysis. Fatty acid methylesters were extracted and resolved by TLC as described in SI Appendix. P. berghei Biotin Ligase KOs. To generate the double cross-over KO construct for PbHCS1 and PbHCS2, homology arms composed of the 5 and 3 UTR of PbHCS1 and PbHCS2 were cloned into the pdef-hdhfr plasmid (51). Constructs were linearized and transfected into P. berghei as described by Janse et al. (49) with a few modifications (SI Appendix). P. berghei Liver-Stage Development and Infectivity Assays. For prepatency assays, salivary gland sporozoites were harvested from infected A. stephensi mosquitoes and injected into mice i.v. followed by daily monitoring of E Dellibovi-Ragheb et al.